I have a graph with three overlapping spectra and I can distinguish them by eye but actually automating their being separated and analysed individually isn’t going super well. If anyone can help with this, I’d also be very grateful!
do not attempt to turn an illustration back into data. contact the authors of the program that makes your plot. ask them how to get the data.
at least show your plots. someone asking to see those plots was inevitable.
Yeah that’s fair. I generated the plots myself and can get the information I need without going via the plot, I was just curious to see whether there was a way to make it work as I figured that may be useful for other applications.
I tried to upload a figure before but I don’t think it worked, let’s see if this works…
I’ll admit this is a particularly messy example too, which won’t help, but I was just curious to see whether the principle of identifying overlapping spectra from either a heatmap (as above) or a scatter plot would be viable, and whether it is something others have done before.
Is it really image processing?
For audio it"s called blind source separation
or even Independent Component Analysis
also, if you’re curious :
// An Introduction to Independent Component Analysis: InfoMax and FastICA algorithms // Dominic Langlois, Sylvain Chartier, and Dominique Gosselin University of Ottawa // Tutorials in Quantitative Methods for Psychology 2010, Vol. 6(1), p. 31-38. //
(same idea for images)